linux-fsdevel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Jens Axboe <jens.axboe@oracle.com>
To: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Miklos Szeredi <miklos@szeredi.hu>,
	linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org,
	linux-mm@kvack.org, torvalds@linux-foundation.org
Subject: Re: [rfc patch 3/4] splice: remove confirm from pipe_buf_operations
Date: Tue, 24 Jun 2008 14:22:58 +0200	[thread overview]
Message-ID: <20080624122258.GR20851@kernel.dk> (raw)
In-Reply-To: <200806242216.41548.nickpiggin@yahoo.com.au>

On Tue, Jun 24 2008, Nick Piggin wrote:
> On Tuesday 24 June 2008 21:36, Miklos Szeredi wrote:
> > > It's an unfortunate side effect of the read-ahead, I'd much rather just
> > > get rid of that. It _should_ behave like the non-ra case, when a page is
> > > added it merely has IO started on it. So we want to have that be
> > > something like
> > >
> > >         if (!PageUptodate(page) && !PageInFlight(page))
> > >                 ...
> > >
> > > basically like PageWriteback(), but for read-in.
> >
> > OK it could be done, possibly at great pain.  But why is it important?
> 
> It has been considered, but adding atomic operations on these paths
> always really hurts. Adding something like this would basically be
> another at least 2 atomic operations that can never be removed again...
> 
> Provided that you've done the sync readahead earlier, it presumably
> should be a very rare case to have to start new IO in the loop
> below, right? In which case, I wonder if we couldn't move that 2nd
> loop out of generic_file_splice_read and into
> page_cache_pipe_buf_confirm. 

That's a good point, moving those blocks of code to the other end makes
a lot of sense. Or just kill the read-ahead, or at least do it
differently. It's definitely an oversight/bug having splice from file
block on the pages it just issued read-ahead for.

> > What's the use case where it matters that splice-in should not block
> > on the read?
> 
> It just makes it generally less able to pipeline IO and computation,
> doesn't it?

Precisely!

> > And note, after the pipe is full it will block no matter what, since
> > the consumer will have to wait until the page is brought uptodate, and
> > can only then commence with getting the data out from the pipe.
> 
> True, but (especially with patches to variably size the pipe buffer)
> I imagine programs could be designed fairly carefully to the size of
> the buffer (and not just things that blast bulk data down the pipe...)

Yep, that's the whole premise for the dynpipe branch I've been carrying
around for some time.

-- 
Jens Axboe


  reply	other threads:[~2008-06-24 12:23 UTC|newest]

Thread overview: 38+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-06-21 15:46 [rfc patch 0/4] splice: cleanups and fixes Miklos Szeredi
2008-06-21 15:46 ` [rfc patch 1/4] splice: fix comment Miklos Szeredi
2008-06-21 15:46 ` [rfc patch 2/4] splice: remove steal from pipe_buf_operations Miklos Szeredi
2008-06-24  8:01   ` Jens Axboe
2008-06-24  8:50     ` Miklos Szeredi
2008-06-24 12:21       ` Nick Piggin
2008-06-21 15:46 ` [rfc patch 3/4] splice: remove confirm " Miklos Szeredi
2008-06-24  8:04   ` Jens Axboe
2008-06-24  8:54     ` Miklos Szeredi
2008-06-24 11:19       ` Jens Axboe
2008-06-24 11:36         ` Miklos Szeredi
2008-06-24 11:46           ` Evgeniy Polyakov
2008-06-24 12:02             ` Miklos Szeredi
2008-06-24 12:15               ` Evgeniy Polyakov
2008-06-24 12:16           ` Nick Piggin
2008-06-24 12:22             ` Jens Axboe [this message]
2008-06-24 13:00             ` Miklos Szeredi
2008-06-24 17:30           ` Linus Torvalds
2008-06-24 18:24             ` Miklos Szeredi
2008-06-24 18:31               ` Linus Torvalds
2008-06-24 19:05                 ` Miklos Szeredi
2008-06-24 19:17                   ` Linus Torvalds
2008-06-24 19:24                     ` Miklos Szeredi
2008-06-24 19:26                       ` Miklos Szeredi
2008-06-24 19:32                         ` Miklos Szeredi
2008-06-24 19:47                           ` Linus Torvalds
2008-06-24 20:06                             ` Miklos Szeredi
2008-06-24 19:45                       ` Linus Torvalds
2008-06-21 15:46 ` [rfc patch 4/4] splice: use do_generic_file_read() Miklos Szeredi
2008-06-24  8:05   ` Jens Axboe
2008-06-24 11:11     ` Miklos Szeredi
2008-06-21 17:20 ` [rfc patch 0/4] splice: cleanups and fixes Subrata Modak
2008-06-22  6:16   ` Miklos Szeredi
2008-06-23 15:26     ` Subrata Modak
2008-06-25 13:17       ` [LTP] " Subrata Modak
2008-06-25 14:00         ` Miklos Szeredi
     [not found]           ` <E1KBVXV-000618-C9-8f8m9JG5TPIdUIPVzhDTVZP2KDSNp7ea@public.gmane.org>
2008-06-26  7:34             ` Subrata Modak
2008-10-22 12:41               ` [LTP] " Subrata Modak

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20080624122258.GR20851@kernel.dk \
    --to=jens.axboe@oracle.com \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=miklos@szeredi.hu \
    --cc=nickpiggin@yahoo.com.au \
    --cc=torvalds@linux-foundation.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).